Inflammatory burden, lifestyle and atherosclerotic cardiovascular disease: insights from a population based cohort study

The inflammatory burden as measured by high-sensitivity C-reactive Protein (hsCRP) is recognized as a cardiovascular risk factor, which can however be affected by lifestyle-related risk factors (LRF). Up-to-date the interplay between hsCRP, LRF and presence and extent of atherosclerotic disease is still largely unknown, which we therefore sought to investigate in a contemporary population-based cohort. We included participants from the cross-sectional population-based Hamburg City Health Study. Affected vascular beds were defined as coronary, peripheral, and cerebrovascular arteries. LRF considered were lack of physical activity, overweight, active smoking and poor adherence to a Mediterranean diet. We computed multivariable analyses with hsCRP as the dependent variable and LRF as covariates according to the number of vascular beds affected. In the 6765 individuals available for analysis, we found a stepwise increase of hsCRP concentration both according to the number of LRF present as well as the number of vascular beds affected. Adjusted regression analyses showed an independent association between increasing numbers of LRF with hsCRP levels across the extent of atherosclerosis. We demonstrate increasing hsCRP concentrations according to both the number of LRF as well as the extent of atherosclerosis, emphasizing the necessity of lifestyle-related risk factor optimization.

Table 1.Baseline characteristics of the total study population and according to the number of affected vascular systems.Missing data of the variables needed for regression analysis and for the classification of subgroups were handled through multivariate imputation by chained equations (MICE).Categorical variables are shown as absolute numbers and percentages, comparison between subgroups was made using the Chisquared test.Continuous variables are described by median and the 1st/3rd quartile, comparison between subgroups was made using the Kruskall-Wallis test.CAD coronary artery disease; CeVD Cerebrovascular disease; HbA1C Glycated haemoglobin; HDL-C high-density lipoprotein cholesterol; hsCRP high-sensitivity C-reactive protein; LDL-C low-density lipoprotein cholesterol; PAD Peripheral artery disease.

Association of hsCRP with LRF
In unadjusted analysis, the number of LRF showed a strong association with hsCRP both in the total cohort as well as according to the number of affected vascular beds (see Table S7).After adjusting for confounders, the association of LRF with hsCRP proved to be consistent in the overall cohort (1 LRF: Beta 0. ; all p < 0.001) after multivariable adjustment.Other LRF such as smoking showed a consistent and significant association with hsCRP across the extent of ASCVD, whilst the association of PA, and sMDS with hsCRP varied (see Table S9, and furthermore Table S10 for analyses without multiple imputations).

Discussion
The current study investigating the association of hsCRP with LRF stratified by the presence and extent of ASCVD in this large-scale contemporary population-based cohort yielded the following main findings: 1.An increase in hsCRP concentrations was found with both an increasing number of LRF and with a greater extent of ASCVD.www.nature.com/scientificreports/www.nature.com/scientificreports/ 2. The highest concentration of hsCRP was found in individuals with the greatest atherosclerotic burden and most LRF, whilst the lowest concentrations were documented in patients without ASCVD and LRF. 3.Even after controlling for major confounders a significant and independent relationship between number of LRF and hsCRP levels was noted, with increasing trend across the extent of ASCVD.
Overall, this underscores the strong association between LRF, low-grade inflammation quantified by hsCRP, and the presence and extent of ASCVD.
Our study represents a contemporary central European population-based cohort with a high rate of ASCVD, since approximately half of the total sample studied presented with atherosclerosis in either the coronary, peripheral or cerebral arteries.With 6.7% the prevalence of CAD in our cohort is comparable to other Western European large scale studies, such as the UK Biobank, where 4.6% of all participants reported atherosclerotic coronary changes 15 .However, a strikingly high rate of PAD with 20.5% and CeVD in 35.8% of the cohort was determined.This is in concurrence with a previously published investigation of the overall HCHS cohort (n = 10,000 participants), where PAD and CeVD were diagnosed in a similar frequency (23.6% and 30.2%, respectively) 16 .In our study, CeVD was defined either by medical history or through imaging findings on carotid ultrasound incorporating intima media thickness (IMT), plaques and carotid stenosis.This in contrast to other populationbased cohorts such as the Study of Health in Pomerania, where prevalent ASCVD was registered by participant history only (20% of the overall cohort at baseline) 17 .A recent meta-analysis investigating the prevalence of CeVD (investigating IMT and hemodynamically relevant stenosis of the carotid arteries, respectively) among individuals from 30 to 79 years of age is in line with our findings.Here, a carotid stenosis was diagnosed in 1.5% of the study population, whilst in 24.4% of all investigated individuals either a carotid plaque or increased IMT was registered 18 .
With regard to lower extremity PAD prevalence our findings are in agreement with other studies, albeit a varying prevalence has been reported on a population-based level in the literature (e.g.21% pathological ABI in the German Epidemiological Trial on Ankle Brachial Index study and 17.8% in the Prevalence of Peripheral Arterial Disease in Subjects with a Moderate Risk of Cardiovascular Disease in Primary Prevention population) 19,20 .In our study both imaging and non-invasive tests such as ABI measurement were used to diagnose PAD.Hence the relatively increased proportions of CeVD and lower extremity PAD are most likely due to the use of non-invasive tests, which identified subclinical atherosclerosis.In summary our study paints a realistic picture of prevalent ASCVD in a European middle-aged population-based sample, underpinning the essential role of screening methods for the early identification of atherosclerotic changes within the arterial vascular tree.
Concerning the overall inflammatory burden in our study, we report low concentrations of hsCRP with a median of 1.1 mg/l in the total study cohort.Our findings are corroborated by reports from other populationbased cohorts.Steppuhn and colleagues reported a median hsCRP level of 1.15 mg/l from the German Health Interview and Examination Survey for Adults investigating 7006 adults aged 18 to 79 years 21 .Slightly higher hsCRP concentrations of 1.5 mg/l were present in the European prospective investigation into cancer in Norfolk population study (EPIC) 22 .Moreover, in diseased cohorts such as the INTERCATH study a median hsCRP of 1.8 mg/l was noted in a cohort of 1014 CAD patients 14 .Whilst the inflammatory burden in our overall cohort was low, we were able to demonstrate that both a higher burden of LRF and a greater extent of ASCVD is associated with rising hsCRP concentrations.This independent association of the number of LRF with hsCRP concentrations was seen across the strata of ASCVD.Our findings have both clinical and scientific implications.Recently, Ridker and colleagues were able to demonstrate that in patients already treated with statins the inflammatory burden as measured by hsCRP was the strongest predictor for major adverse cardiovascular events, and both cardiovascular as well as all-cause mortality 7 .In a study from Blaum and colleagues, the investigators were able to demonstrate that the number of LRF was strongly associated with hsCRP concentrations.Furthermore, and similar to our results, the authors demonstrated in their CAD cohort that overweight had the strongest association with the inflammatory burden 14 .This finding has been validated in multiple studies where adiposity was shown to correlate with an inflammatory state, potentially through secretion of pro-inflammatory cytokines such as IL-6 or TNF-a by adipose tissue 13,23 .Accordingly, a positive effect on the inflammatory load after weight loss has been demonstrated in overweight patients 24,25 .The anti-inflammatory potential inherent in risk factor optimization is underlined in the study by Blaum and colleagues, where 37.9% of the study cohort would achieve hsCRP levels < 2 mg/l after a hypothesized optimization of risk factors (i.e.weight loss, smoking cessation, adherence to a mediterranean diet, regular PA), and therefore to levels below the threshold used in the enrolment of seminal trials investigating anti-inflammatory treatment (i.e.CANTOS and COLCOT) 3,4,14 .Thus, solely by the optimization of their LRF burden, a relevant health benefit could be unlocked by the patients themselves ahead of the initiation of a specific anti-inflammatory therapy and the potentially associated side-effects of these medications.Our findings therefore underline the necessity of lifestyle optimization as recommended in the guidelines on cardiovascular disease prevention in clinical practice from the European Society of Cardiology 26 .
In our cohort individuals with the greatest atherosclerotic burden (i.e.≥ 2 vascular systems affected) and most LRF had the highest concentrations of hsCRP, displaying the synergistic effect of atherosclerosis itself and LRF on the overall inflammatory burden.Patients with atherosclerosis in two or more vascular beds are a sub-population in whom a particularly high risk for recurrent events has been described 27 .An intensified approach to secondary prevention has therefore been proposed to reduce the incidence of adverse cardiovascular events and mortality in this patient population 28 .Accordingly, optimization of LRF should particularly be prioritized in this cohort.
On the other end of the spectrum, we were also able to show that in participants without ASCVD, the number of LRF was significantly and independently associated with hsCRP concentrations.Besides the use of CRP as biomarker to assess the residual inflammatory risk, the causal role of this acute-phase protein in the development of ASCVD has also been investigated 9 .A contribution of CRP to endothelial dysfunction and hypertension by the inhibition of nitric oxide, impaired endothelial-associated vascular relaxation, and association with plaque instability by activating NF-κB has been demonstrated 29 .Therefore, it can be hypothesized that the optimization of LRF might be associated with a delayed development of atherosclerotic precursors.However, further studies are needed to ascertain these findings in primary prevention.

Limitations
Whilst the studied population represents a large contemporary cohort, some limitations merit consideration.Since random recruitment took place from a statistical sample of a middle-aged population of the German city of Hamburg, a translation of our findings to other ethnicities, age groups, or geographical regions can only be carried out with caution.Moreover, whilst the selection of participants was carried out randomly, a recruitment bias is possible since healthier individuals are more likely to accept taking part in a study.Also, a relevant proportion of participants (32%) of the overall cohort of 10,000 individuals had to be excluded due to missing hsCRP samples, and comorbidities such as prevalent neoplastic diseases or inflammatory disorders.Lastly, whilst CeVD and PAD were deemed present either by medical history or imaging studies, for CAD solely self-reported information from the individual participants' medical history was available.However, the CAD prevalence reported in our study is comparable to other large-scale population-based cohorts such as the UK Biobank.

Conclusion
In this contemporary population-based cohort, a significant association of lifestyle-related risk factors such as diet, physical activity, smoking, and overweight with the inflammatory burden as measured by hsCRP across the extent of atherosclerotic disease was determined.We demonstrate that both the presence and extent of ACSVD and the burden of LRF have additive effects on the inflammatory load.These findings emphasize the important role of subclinical inflammation in individuals with and without ASCVD and might be helpful for the definition of target populations for anti-inflammatory compounds across the extent of atherosclerotic disease.

Study cohort
The Hamburg City Health Study (HCHS) is a prospective population-based cohort study that aims to gather extensive data on risk factors for chronic diseases, with a focus on organ system function and structure through a wide range of assessments.It involves 45,000 randomly selected participants from Hamburg, Germany, aged 45 to 74, and includes a detailed examination of lifestyle, environment, genetics, and health outcomes.This large-scale, long-term assessment, conducted in a European metropolitan population, seeks to explore the interplay between biological and psychosocial factors in the context of health and disease.Hereby, the HCHS aims to identify risk factors for major chronic diseases and survivorship, with the goal of accurately determining the prevalence and incidence of common diseases, and the development of complex models for predicting health outcomes.More details on the study are described elsewhere 30 .
The current cross-sectional analysis incorporates data from the first 10,000 individuals.Baseline examinations, which included the completion of validated patient questionnaires to self-report lifestyle, blood draws, and non-invasive tests, were conducted during a single-day visit at the HCHS Epidemiological Study Center of the University Medical Center Hamburg-Eppendorf between 2016 and 2018.An ethics approval was obtained from the ethics committee of the Medical Association of Hamburg (PV5131) and the study is registered at ClinicalTrials.gov(NCT03934957).Written informed consent was obtained from all participants.Moreover, the study and all applied methods were performed in accordance with the relevant guidelines and regulations.

In-and exclusion criteria
Participants with valid measurements of hsCRP (i.e.within the limit of detection of the used assay) concentration at baseline were included in the current study.Individuals with chronic inflammatory disorders (defined as rheumatoid arthritis, multiple sclerosis, psoriasis, inflammatory bowel disease), prevalent cancer, intake of immunosuppressive or antineoplastic medication, or missing information of more than 3 values of the previously mentioned criteria were excluded.Also, participants with missing hsCRP concentrations or hsCRP concentrations > 10 mg/l compatible with an acute infection were excluded.After applying the in-and exclusion criteria 6675 datasets were left for the presented analyses (see Fig. 2 for further details).The reporting of the current study was in accordance with the Strengthening the reporting of observational studies in epidemiology (STROBE) statement 31 .

Assessment of cardiovascular risk factors, co-morbidities and further variables
Intake of medications was validated via Anatomical Therapeutic Chemical Classification System Codes (ATC-Code) for anti-hypertensive, anti-diabetic medications and statins.Diabetes mellitus was defined as taking medication of the ATC group A10, a fasting glucose > 126 mg/dL, a non-fasting glucose > 200 mg/dL, an HbA1C > 6.5%, or self-reported diabetes.Arterial hypertension was considered to be prevalent when taking medication in ATC group C09A, C09C, C07A, C03C, C03A, C03D, C08C, C02D, C02A, C09X, C01D, a measured resting systolic blood pressure > 140 mmHg/resting diastolic blood pressure > 90 mmHg at inclusion, or selfreported hypertension.Prevalent chronic kidney disease (CKD) was defined as estimated glomerular filtration rate (eGFR) below 60 ml/min as estimated by the CKD-EPI formula or by medical history 32  www.nature.com/scientificreports/

Laboratory measurements
Laboratory measurements (including quantification of glycated hemoglobin (HbA1C), cholesterol levels and hsCRP) were carried out at the HCHS laboratories on the same day as the baseline examination.HsCRP was quantified using a commonly available assay (SIEMENS Healthineers, Atellica High Sensitivity C-reactive protein; Range: 0.16-10.00mg/l, limit of detection ≤ 0.16 mg/l).Calibration was carried out within the clinical routine.

Assessment of atherosclerotic burden
Prevalent CAD was assessed via self-reported medical history.CeVD was diagnosed by medical history including previous stroke, or by IMT of ≥ 1 mm, vascular plaque (defined as localized IMT thickening of ≥ 1.5 mm), or stenoses (defined as systolic flow velocity > 200 m/s in the common, internal and external carotid artery) on carotid ultrasound at inclusion 33 .PAD was diagnosed via medical history, or pathological ankle-brachial-index (ABI, ≤ 0.9) 34 .Proportions of patients with history of ASCVD and atherosclerosis detected at baseline (for CeVD and PAD) are shown in Table S1.

Assessment of lifestyle-related risk factors
LRF that were taken into consideration were lack of physical activity (PA), defined as < 1.5 h/week of exercise, overweight, defined as BMI ≥ 25 kg/m 2 , current smoking, defined as active smoking or recently quitted smoking within the last 6 months, and poor adherence to a Mediterranean diet.A simple Mediterranean diet score (sMDS) as used in the Stabilisation of atherosclerotic plaque by initiation of darapladib therapy (STABILITY) trial was calculated to investigate dietary habits 35 .Briefly, participants answered a food questionnaire in which 4 food groups (consumption of fruit, vegetables, fish, and alcohol) were queried, and points according to the frequency of consumption (maximum 2 points, minimum 0 points) were distributed (see supplementary Table S3).
A score ranging from 0 to 8 points was then calculated.Poor adherence to Mediterranean diet was defined as an sMDS ≤ 2 points.

Statistical analyses
Categorical variables are shown as absolute numbers and percentages, compared using Chi-squared test.Continuous variables are reported as median and 1st/3rd quartile (Q1/3) and were compared using Kruskall-Wallis test.
To analyse the burden of ASCVD, the population was divided into subgroups according to the number of vascular beds affected (no ASCVD, 1 affected vascular system, ≥ 2 affected vascular systems).Uni-and multivariable linear regression models in the overall cohort and each subgroup with logarithmic hsCRP as dependent variable and LRF (both number of LRF as a categorical variable and each individual LRF) as covariate were calculated and adjusted for age, sex, diabetes, arterial hypertension, intake of statins and chronic kidney disease.Missing data of the variables needed for regression analysis and for the classification of subgroups (overweight, PA, sMDS, smoking, diabetes, CAD, PAD and CeVD) were handled by Multivariate Imputation by Chained Equations (MICE), as proposed by Buuren and Groothuis-Oudshoorn (20 imputed data sets, R Package MICE) 36 .Results were considered as statistically significant at a significance level of a two-sided alpha < 0.05.All statistical tests were carried out using R version 4.1.2.

Previous Presentation
Part of this work has been presented as an abstract presentation at the Congress of the European Society of Cardiology 2023.

Figure 1 .
Figure 1.Median high-sensitivity C-reactive protein levels according to the number of affected vascular systems and number of lifestyle-related risk factors.HsCRP High-sensitivity C-reactive protein; LRF Lifestylerelated risk factors.

Table 2 .
Lifestyle-related risk factors of the total study population and according to the number of affected vascular systems.Missing data of the variables needed for regression analysis and for the classification of subgroups were handled through multivariate imputation by chained equations (MICE).Categorical variables are shown as absolute numbers and percentages, comparison between subgroups was made using the Chisquared test.Continuous variables are described by median and the 1st/3rd quartile, comparison between subgroups was made using the Kruskall-Wallis test.BMI body mass index; LRF lifestyle-related risk factors; sMDS simple Mediterranean diet score.